Speaking in the opening keynote of SIIA OnDemand in San Jose this morning, SuccessFactors CEO Lars Dalgaard let slip a statistic that set several attendees a-twittering. He revealed that the SaaS provider's multi-tenant application infrastructure supports its 2,850+ customers and 5.4+ million users on just 150 servers.
The ability to achieve such enormous economies of scale demonstrates the huge power of multi-tenancy and gives the lie to the line, so often peddled by the conventional on-premise software vendors, that SaaS is just a delivery option. SuccessFactors would not be able to run its operations with anything like the same low overhead if it had to separately maintain the ability to ship on-premise instances of its software.
The conventional software model perpetuates a scandalous wastage and duplication of resources. Every single customer of an on-premise platform or application installs their own custom implementation. Every one of those implementations builds in enough spare capacity to support unexpected usage spikes and peak load at the organization's busiest period of the year — yet remains idle the rest of the time. Its IT staff acquire a huge store of learnings and experiences that are solely revelant to their own environment. All of those needless investments and expenses are replicated across thousands of an ISV's customer base. The aggregate waste adds up to a burdensome cost of ownership spread across its customer.
With SaaS, the customer base shares a single infrastructure, eliminating all of the redundant capacity and wasted effort of the on-premise model. The aggregate cost of ownership is dramatically reduced, as illustrated in the diagram above. Those lower costs translate into savings for customers as well as better margins for providers. With cloud platforms and PaaS, where ISVs share a common infrastructure, even more waste and cost is squeezed out.
These savings aren't just a one-time phenomenon. Having all its customers on a single, shared infrastructure allows a SaaS ISV or cloud provider to continually analyze usage and performance so that it can carry on innovating to enhance the value it delivers. Salesforce.com recently announced it is moving to a maximum five minutes of customer downtime for scheduled upgrades. I met today with LiveOps, a provider of on-demand call center services, which manages its upgrades in a way that means customers suffer no downtime at all. It achieves a stellar four nines of uptime including scheduled maintenance. And of course this is within infrastructures whose security and firewall strength is far superior to those of most enterprise data centers — and which is constantly tested and validated by customers and prospects to an extent no single enterprise data center experiences.
This is what the cloud delivers: hardened, high-performance, reliable infrastructure far superior to that of any single enterprise — at an aggregate cost of ownership and operation far lower than those enterprises would collectively spend across hundreds or thousands of separate private infrastructures to achieve a much worse result.
Disclosure: the diagram shown is excerpted from Redefining software platforms — How PaaS changes the game for ISVs, an analyst report produced with funding from Intuit Partner Platform.